Predicting a Salient Social Identity from Linguistic Style

Abstract

Social media data are already being used to classify individuals into mutually exclusive social groups. Here we propose a model based on self-categorization theory that classifies which of two social identities is salient within the same person using text data. Based on over 500,000 online forum posts and seven prototype-based style features, a trained classifier correctly distinguishes between posts written by the same person in two different social contexts – a parenting forum and a feminist forum – significantly above chance level (AUC = .74). We then apply the trained classifier to a new dataset (N = 153) obtained from an online experiment where salience of group membership is manipulated. We show that our trained model distinguishes between salient parent and feminist identities significantly above chance level when the topic is irrelevant to either identity (AUC = .69). We discuss applications but also limitations of a text-based prediction of salient social identities.

Using computational techniques to study social influence online

“…Further, Chung and Pennebaker (2019), the developers of LIWC, advise using a minimum cut-off of 100 words where possible. In practice, word count cut-offs are often lower than this, especially when using sparser social media data (50 words – Bäck et al, 2018; 25 words – Koschate et al, 2019; 45 words – Nelson et al, 2017; 50 words – Petrie et al, 2008; 50 words – Wilson, 2019). Our choice of 50 words was made in order to keep as much data as possible in our analysis, whilst ensuring that the data could be used to draw psycho-logically meaningful conclusions (Pennebaker Conglomerates Inc., 2017).…”

“…We first collected data from individuals writing with one of our two identities salient. In line with Koschate et al (2019), we used forum topic as a proxy for identity salience. We assumed that individuals posting in either an entrepreneur forum or a libertarian forum had the respective identities salient at the time of writing.…”

“…It therefore follows that, based on the idea that individuals are motivated to communicate their social identities online, it will be possible to detect salient social identities using online behavioral data. In research by Koschate et al (2019), the prototypical linguistic style of group members is used to detect the salience of either a parent or feminist identity. Whilst using forum as a proxy for identity salience, they demonstrate that it is possible to predict which forum a post originated in based only on linguistic style features.…”

“…In Study 1a, we trained and validated a decision tree classifier to detect our two social identities – a libertarian identity and an entrepreneur identity – using Reddit data. We hypothesized that, in line with Koschate et al (2019), it would be possible to accurately differentiate between two social groups based on a group prototypical linguistic style (H1). We then validated our model by testing (1b) whether it could correctly classify posts where the author is the same person (within-person design), and (1c) whether this finding is more than accommodation to local forum norms using data from Silk Road.…”

“…One of the primary barriers to studying group processes online from a sociopsychological perspective has been the lack of methodology for measuring social identity constructs without utilizing self-report survey data. In this article, we build upon a novel method by Koschate et al (2019) to assess group prototypicality in online text data. This allows us to test whether prototypicality relates to social influence in online forums using naturally occurring behavioral data such as linguistic and social network indicators rather than self-report measures.…”